import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.dates as mdate
from datetime import datetime
from statistics import median

'''
IMPORTANT: Remove leading space from column '# Bytes' in csv file or pandas
won't read it correctly.
'''

dataset = pd.read_csv('./global_last10years.csv', index_col=0).fillna(0)

# Convert unix timestamps to dates
timestamps = mdate.epoch2num(dataset.index)
date_formatter = mdate.DateFormatter('%y-%m-%d %H:%M:%S')

# List of packet counts per hour (daily avg)
ts_packets = dataset.iloc[:,1]

# List of bytes per hour (daily avg)
ts_bytes = dataset.iloc[:,0]

# List of unique source IPs per hour (daily avg)
ts_uIPs = dataset.iloc[:,2]

# List of unique destination IPs per hour (daily avg)
ts_uIPd = dataset.iloc[:,3]

### rep-14: Signals correlation ###

print('Correlations between different columns (NaNs replace with 0):')
print(dataset.corr(method='pearson').round(decimals=2))

### rep-15: Destinations/Sources ratio ###

ratio_uIPd_uIPs = median(ts_uIPd) / median(ts_uIPs)
print('Ratio between median IP destinations and median IP sources: ', ratio_uIPd_uIPs)

### rep-16: Peak in unique sources ###

date = datetime.fromtimestamp(ts_uIPs.idxmax())
print('Peak in unique sources: ', date)